| 1 |
<?php |
| 2 |
|
| 3 |
namespace ABlocksLinkGuard; |
| 4 |
|
| 5 |
if ( ! defined( 'ABSPATH' ) ) { |
| 6 |
exit; // Exit if accessed directly. |
| 7 |
} |
| 8 |
|
| 9 |
/** |
| 10 |
* Maps an `href` to the post it points at. |
| 11 |
* |
| 12 |
* `url_to_postid()` understands every permalink structure, custom post type |
| 13 |
* and page hierarchy, but it answers the question "can the *current visitor* |
| 14 |
* see this?" rather than "which post is this?". A scheduled post resolves for |
| 15 |
* an editor and resolves to 0 for everyone else, because `WP_Query` drops |
| 16 |
* unpublished singular results after fetching them. A cache built on that |
| 17 |
* answer would depend on who happened to load the page first. |
| 18 |
* |
| 19 |
* The fetch happens before the drop, and `posts_results` fires in between. So |
| 20 |
* the resolver listens there for the duration of one `url_to_postid()` call and |
| 21 |
* keeps the post it saw, which makes the answer the same for every viewer. |
| 22 |
*/ |
| 23 |
class Resolver { |
| 24 |
|
| 25 |
/** |
| 26 |
* Post seen by the singular query during the current resolve() call. |
| 27 |
* |
| 28 |
* @var int |
| 29 |
*/ |
| 30 |
private static $captured = 0; |
| 31 |
|
| 32 |
/** |
| 33 |
* Slug the singular query asked for during the current resolve() call. |
| 34 |
* |
| 35 |
* @var string |
| 36 |
*/ |
| 37 |
private static $slug = ''; |
| 38 |
|
| 39 |
/** |
| 40 |
* Reduce an href to a canonical same-site URL, or '' when it cannot be a |
| 41 |
* link to a post (another site, an anchor, mailto:, an upload, wp-admin...). |
| 42 |
* |
| 43 |
* The result is also the cache key, so two spellings of the same link — |
| 44 |
* relative and absolute, with or without `www.`, a `#section` or a |
| 45 |
* tracking query string — collapse to one entry. |
| 46 |
* |
| 47 |
* @param string $href Raw attribute value. |
| 48 |
* @return string |
| 49 |
*/ |
| 50 |
public static function normalize( $href ) { |
| 51 |
$href = trim( (string) $href ); |
| 52 |
|
| 53 |
if ( '' === $href || '#' === $href[0] ) { |
| 54 |
return ''; |
| 55 |
} |
| 56 |
|
| 57 |
if ( 0 === strpos( $href, '//' ) ) { |
| 58 |
$href = 'https:' . $href; |
| 59 |
} |
| 60 |
|
| 61 |
$parts = wp_parse_url( $href ); |
| 62 |
if ( false === $parts ) { |
| 63 |
return ''; |
| 64 |
} |
| 65 |
|
| 66 |
if ( isset( $parts['scheme'] ) && ! in_array( strtolower( $parts['scheme'] ), [ 'http', 'https' ], true ) ) { |
| 67 |
return ''; |
| 68 |
} |
| 69 |
|
| 70 |
$home = wp_parse_url( home_url() ); |
| 71 |
|
| 72 |
if ( isset( $parts['host'] ) ) { |
| 73 |
if ( self::bare_host( $parts['host'] ) !== self::bare_host( isset( $home['host'] ) ? $home['host'] : '' ) ) { |
| 74 |
return ''; |
| 75 |
} |
| 76 |
} elseif ( empty( $parts['path'] ) || '/' !== $parts['path'][0] ) { |
| 77 |
// Document-relative links ("next-post/", "?p=3") depend on the page |
| 78 |
// they sit on; guessing their target could hide a working link. |
| 79 |
return ''; |
| 80 |
} |
| 81 |
|
| 82 |
$path = isset( $parts['path'] ) ? $parts['path'] : '/'; |
| 83 |
$home_path = isset( $home['path'] ) ? rtrim( $home['path'], '/' ) : ''; |
| 84 |
|
| 85 |
if ( '' !== $home_path && 0 !== strpos( $path, $home_path ) ) { |
| 86 |
return ''; |
| 87 |
} |
| 88 |
|
| 89 |
$relative = '/' . ltrim( substr( $path, strlen( $home_path ) ), '/' ); |
| 90 |
|
| 91 |
if ( preg_match( '#^/(?:wp-admin|wp-content|wp-includes|wp-json)(?:/|$)|^/wp-[a-z-]+\.php#i', $relative ) ) { |
| 92 |
return ''; |
| 93 |
} |
| 94 |
|
| 95 |
// A file: an upload, a sitemap, a download. Never a post. |
| 96 |
if ( preg_match( '#\.(?!html?$|php$)[a-z0-9]{2,5}$#i', $relative ) ) { |
| 97 |
return ''; |
| 98 |
} |
| 99 |
|
| 100 |
// Keep the query string only when it is what identifies the post. |
| 101 |
$query = ''; |
| 102 |
if ( ! empty( $parts['query'] ) && preg_match( '#(?:^|&)(p|page_id|preview_id)=(\d+)#', $parts['query'], $match ) ) { |
| 103 |
$query = '?' . ( 'page_id' === $match[1] ? 'page_id' : 'p' ) . '=' . $match[2]; |
| 104 |
} |
| 105 |
|
| 106 |
return untrailingslashit( home_url( $relative ) ) . $query; |
| 107 |
} |
| 108 |
|
| 109 |
/** |
| 110 |
* Post ID a normalized URL points at, whatever that post's status. 0 when |
| 111 |
* the URL is not a single post (an archive, a custom route, nothing). |
| 112 |
* |
| 113 |
* @param string $url Output of normalize(). |
| 114 |
* @return int |
| 115 |
*/ |
| 116 |
public static function resolve( $url ) { |
| 117 |
if ( preg_match( '#\?(?:p|page_id)=(\d+)$#', $url, $match ) ) { |
| 118 |
return (int) $match[1]; |
| 119 |
} |
| 120 |
|
| 121 |
self::$captured = 0; |
| 122 |
self::$slug = ''; |
| 123 |
add_action( 'parse_query', [ __CLASS__, 'capture_slug' ], PHP_INT_MAX ); |
| 124 |
add_filter( 'posts_results', [ __CLASS__, 'capture' ], PHP_INT_MAX, 2 ); |
| 125 |
|
| 126 |
try { |
| 127 |
$id = (int) url_to_postid( trailingslashit( $url ) ); |
| 128 |
} finally { |
| 129 |
remove_action( 'parse_query', [ __CLASS__, 'capture_slug' ], PHP_INT_MAX ); |
| 130 |
remove_filter( 'posts_results', [ __CLASS__, 'capture' ], PHP_INT_MAX ); |
| 131 |
} |
| 132 |
|
| 133 |
if ( $id || self::$captured ) { |
| 134 |
return $id ? $id : self::$captured; |
| 135 |
} |
| 136 |
|
| 137 |
return self::trashed( self::$slug ); |
| 138 |
} |
| 139 |
|
| 140 |
/** |
| 141 |
* Trashing a post renames its slug to `{slug}__trashed`, so its old URL |
| 142 |
* stops matching anything and would otherwise look like a link to nowhere. |
| 143 |
* |
| 144 |
* @param string $slug Slug the URL asked for. |
| 145 |
* @return int |
| 146 |
*/ |
| 147 |
private static function trashed( $slug ) { |
| 148 |
if ( '' === $slug ) { |
| 149 |
return 0; |
| 150 |
} |
| 151 |
|
| 152 |
$ids = get_posts( |
| 153 |
[ |
| 154 |
'name' => $slug . '__trashed', |
| 155 |
'post_type' => 'any', |
| 156 |
'post_status' => 'trash', |
| 157 |
'fields' => 'ids', |
| 158 |
'posts_per_page' => 1, |
| 159 |
'suppress_filters' => true, |
| 160 |
'no_found_rows' => true, |
| 161 |
] |
| 162 |
); |
| 163 |
|
| 164 |
return $ids ? (int) $ids[0] : 0; |
| 165 |
} |
| 166 |
|
| 167 |
/** |
| 168 |
* @internal Hooked only for the duration of resolve(). |
| 169 |
* |
| 170 |
* @param \WP_Query $query The query url_to_postid() is about to run. |
| 171 |
*/ |
| 172 |
public static function capture_slug( $query ) { |
| 173 |
if ( '' !== self::$slug || ! $query->is_singular() ) { |
| 174 |
return; |
| 175 |
} |
| 176 |
$slug = $query->get( 'name' ) ? $query->get( 'name' ) : $query->get( 'pagename' ); |
| 177 |
self::$slug = $slug ? sanitize_title( basename( $slug ) ) : ''; |
| 178 |
} |
| 179 |
|
| 180 |
/** |
| 181 |
* @internal Hooked only for the duration of resolve(). |
| 182 |
* |
| 183 |
* @param \WP_Post[] $posts Raw query results. |
| 184 |
* @param \WP_Query $query The query. |
| 185 |
* @return \WP_Post[] |
| 186 |
*/ |
| 187 |
public static function capture( $posts, $query ) { |
| 188 |
if ( ! self::$captured && ! empty( $posts ) && $query->is_singular() ) { |
| 189 |
self::$captured = (int) ( is_object( $posts[0] ) ? $posts[0]->ID : $posts[0] ); |
| 190 |
} |
| 191 |
return $posts; |
| 192 |
} |
| 193 |
|
| 194 |
private static function bare_host( $host ) { |
| 195 |
return preg_replace( '#^www\.#', '', strtolower( $host ) ); |
| 196 |
} |
| 197 |
} |
| 198 |
|
